Temperature of the Year. 39 



As the principal object in view was to derive a rule for obtain- 

 ing the mean temperature from two or three daily observations, 

 my attention was particularly directed to this subject. After ma- 

 king various combinations of the hourly means, and comparing 

 them with the true mean, the following very simple rule was dis- 

 covered. Divide the meteorological year into quarters, thus : — 

 1st quarter. — November, December, January. 

 2d quarter. — February, March, April. 

 3d quarter. — May, June, July. 

 4th quarter. — August, September, October. 

 The equinoxes and solstices fall about in the middle of the 

 quarters respectively. During the first quarter, record the ther- 

 mometer at 9, A. M. and 6, P. M. 

 2d qr. at 8, A. M. and 6, P. M. 

 3d qr. at 7, A. M. and 6, P. M. 

 4th qr. at 8, A. M. and 6, P. M. 

 The following table exhibits a comparison of the true avera- 

 ges, and those obtained from the hours just named. 



True mean. 9 & 6 



Istqr. 29°.01 28°.90 



8&6 



2d qr. 37°.31 37°.35 



7i&6 



3dqr. 630.38 63°.41 



8&6 



4th qr. 59°.23 59°.41 ' 



Whole year, 47°.23 47°.27 



The numbers in these two columns almost perfectly coincide, 

 and it is probable, that were the experiment to be tried in almost 

 any other year, the agreement would be less exact. But since 

 the annual mean temperature at a given place, as is well known, 

 is nearly the same from year to year, and since the hours of ob- 

 servation in the above system have a symmetrical arrangement 

 with regard to the sun's declination, it is beheved the rule will 

 be nearly accurate every year at this place, and at other places 

 whose latitude does not differ widely from this. There must of 

 course be two points of time in the mean annual day, (one in the 

 ascent of the thermometer, the other in its descent,) at which 

 the mean temperature occurred. They are indicated in the 

 plate by the intersections of the annual curve, and the corres- 



